Posts by biodoc
log in
1) Message boards : Number crunching : optimized applications and anonymous platform? (Message 1543)
Posted 20 Aug 2013 by biodoc
Dagorath and frankhagen - stop offending here.
For clarifiaction, I got the source codes from one member of PNT, they have them optimized for SSE2, SSE3 and AVX, I compiled them and tested. They compute fast and results are good. I was comparing results from those application to our original application and the data are ok.


It's good to know the results from the PNT optimized app are valid.
2) Message boards : News : New optimized versions for SSE3 released (Message 1502)
Posted 17 Aug 2013 by biodoc
Thanks Kyong. Sometimes I think Boinc has a mind of its own. :)
3) Message boards : News : New optimized versions for SSE3 released (Message 1500)
Posted 17 Aug 2013 by biodoc
I'm seeing a mixture of WUs. Some are for the old optimized app and some are for the new sse3 app. Is this normal? Can't the new sse3 app crunch all WUs?
4) Message boards : News : New optimized versions for SSE3 released (Message 1494)
Posted 16 Aug 2013 by biodoc
cool!
5) Message boards : News : New optimized versions have been released (Message 1480)
Posted 11 Aug 2013 by biodoc
I moved my 1090T (64 bit Linux Mint 15) on to the project and it is completing
WUs 2x faster with the new optimized app.

My 2600K (64 bit Linux Mint 15) is "only" 30% faster with the new app.
9907665 3851208 8 Aug 2013, 14:13:00 UTC 9 Aug 2013, 3:18:22 UTC Completed and validated 6,019.49 5,949.34 120.00 Period Search Application v102.00 9903751 3880191 7 Aug 2013, 21:47:18 UTC 8 Aug 2013, 7:59:00 UTC Completed and validated 6,314.24 6,238.79 120.00 Period Search Application v102.00 9902604 3674191 7 Aug 2013, 12:15:50 UTC 7 Aug 2013, 23:44:36 UTC Completed and validated 9,155.71 9,055.44 120.00 Period Search Application v101.00 9901057 3865553 7 Aug 2013, 4:02:30 UTC 7 Aug 2013, 15:58:52 UTC Completed and validated 8,925.42 8,828.53 120.00 Period Search Application v101.00


Nothing was changed before and after (old app vs new app). Hyperthreading was on and I was running GPUgrid.

No big deal, I just thought is was interesting.
6) Message boards : News : New optimized versions have been released (Message 1464)
Posted 9 Aug 2013 by biodoc
HT is on and I'm also running GPU grid on a GTX460. So 8 threads total and GPU grid is stealing processor time both on the old and new asteroids app.

I'll experiment a bit and get back to you.
7) Message boards : News : New optimized versions have been released (Message 1462)
Posted 9 Aug 2013 by biodoc
Any idea why the linux64 app isn't 2x faster? There's an improvement, but not a 2 fold decrease in crunch time like the windows app.

EDIT: link to my 2600K running linux

http://asteroidsathome.net/boinc/results.php?hostid=37254
8) Message boards : Number crunching : optimized applications and anonymous platform? (Message 1399)
Posted 23 Jul 2013 by biodoc
I guess there are 2 important questions. Are the the computers running the optimized application producing scientifically valid results and if so, are the developer(s) willing to share the optimized source code with Kyong?

i'm hoping that those who are currently utilizing optimized applications for this project (if that's in fact what's going on here) aren't PPD zealots. it will ultimately hurt the project if the people with the skills to create optimized apps kept those apps to themselves. of course i don't want to jump to conclusions and offend anyone, so perhaps no optimized apps have been published on the message boards yet b/c these developers haven't yet perfected their optimizations. for instance, while the project's top host (whose owner remains anonymous) has 5278 valids and 0 invalids, a substantial fraction of that host's errors are actual computation errors (as opposed to deadline lapses or server-side cancellations).

i'm still waiting to hear back from Thor.


Only Kyong can confirm whether or not the results the optimized app are producing are scientifically valid or not. I don't trust boinc to declare the results to be valid. I hope I'm wrong though.
9) Message boards : Number crunching : optimized applications and anonymous platform? (Message 1393)
Posted 23 Jul 2013 by biodoc
I guess there are 2 important questions. Are the the computers running the optimized application producing scientifically valid results and if so, are the developer(s) willing to share the optimized source code with Kyong?
10) Message boards : Number crunching : optimized applications and anonymous platform? (Message 1385)
Posted 22 Jul 2013 by biodoc
7 of the top 11 hosts are listed as "owner: anonymous". All are running "Period Search Application Anonymous platform (CPU)". All have quite amazing crunch times.

This one: http://asteroidsathome.net/boinc/results.php?hostid=35494 shows crunch times more than ten fold faster than my 2600K.

Is this some sort of closed beta testing of a new optimized application?




Main page · Your account · Message boards


Copyright © 2021 Asteroids@home